Objective The study aims to validate eye movement tracking performed without a baseline study as a biomarker for concussion in a paediatric population.

Design This is a cross-sectional case control study of patients diagnosed with concussion in a referral centre compared to non-concussed controls.

Methods Eye movements were recorded with an SR Research Eyelink 1000 eye tracker while a 220-second video was played. Eye tracking metrics were compared to Acute Concussion Evaluation (ACE) scores and clinical assessments of convergence and accommodation.

Results 56 children with concussions (mean age of 13 years), as defined by symptom presence following an impulsive injury to the brain were evaluated at a mean of 22 weeks post-injury. Twelve eye tracking metrics were significantly different between concussed children and 83 uninjured controls. A model built on a balanced sub-sample to classify concussion based on eye tracking achieved an AUC of 0.854 (sensitivity 71.9%; specificity 84.4%). Two metrics were significant predictors of abnormal near point convergence (NPC) and accommodation. The model built to classify concussion based on NPC status achieved a specificity of 95.8% and a sensitivity of 57.1%. Reduced binocular amplitude of accommodation had a spearman correlation of 0.752 (p-value<0.001) with NPC. Seven eye tracking metrics were found to have a strong correlation (R=0.781) with binocular amplitude of accommodation.

Conclusions Eye tracking reliably detected concussion and convergence and accommodative abnormalities in the paediatric population. We propose that eye tracking without baseline assessment may serve as an objective measure of concussion in paediatric patients.

Competing interests Uzma Samadani has submitted intellectual property describing the technology utilised in this paper.

These patents are owned by NYU, the VA and HCMC and licensed to Oculogica Inc., a company in which all of the above parties have an equity interest